Self-improvement & synthetic data
56 papers in this thread, across 14 domains.
Progress0 of 17
- 2026Principled Synthetic Data Enables the First Scaling Laws for LLMs in RecommendationLLM Systems2602.07298AI at MetaFeb 7, 2026score 4~120 minLLM Systems
- 2025LaSeR: Reinforcement Learning with Last-Token Self-RewardingRL Training2510.14943Tencent HunyuanOct 16, 2025score 10~101 minRL Training
- 2025Repurposing Synthetic Data for Fine-grained Search Agent SupervisionRL Training2510.24694TongyiLabOct 28, 2025score 9~105 minRL Training
- 2025Beyond Pass@1: Self-Play with Variational Problem Synthesis Sustains RLVRRL Training2508.14029Aug 19, 2025score 9~128 minRL Training
- 2025A Modular Approach for Clinical SLMs Driven by Synthetic Data with Pre-Instruction Tuning, Model Merging, and Clinical-Tasks AlignmentTraining Methods2505.10717Microsoft ResearchMay 15, 2025~132 minTraining Methods
- 2024Source2Synth: Synthetic Data Generation and Curation Grounded in Real Data SourcesData2409.08239Sep 12, 2024score 10~112 minData
- 2024MUTUAL REASONING MAKES SMALLER LLMS STRONGER PROBLEM-SOLVERSReasoning2408.06195Aug 12, 2024~115 minReasoning
- 2024Scaling Synthetic Data Creation with 1,000,000,000 PersonasData2406.20094Jun 28, 2024~88 minData
- 2024Self-Play Preference Optimization for Language Model AlignmentAlignment2405.00675May 1, 2024score 9~96 minAlignment
- 2024DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic DataReasoning2405.14333DeepSeekMay 23, 2024score 9~101 minReasoning
- 2024Self-Play Fine-Tuning Converts Weak Language Models to Strong Language ModelsAlignment2401.01335Jan 2, 2024score 9~128 minAlignment
- 2024Self-Rewarding Language ModelsAlignment2401.10020Jan 18, 2024score 9~116 minAlignment
- 2023ReST meets ReAct: Self-Improvement for Multi-Step Reasoning LLM AgentAgents2312.10003Dec 15, 2023score 9~107 minAgents
- 2023Beyond Human Data: Scaling Self-Training for Problem-Solving with Language ModelsTraining Methods2312.06585Dec 11, 2023score 10~116 minTraining Methods
- 2023Synthetically Enhanced: Unveiling Synthetic Data's Potential in Medical Imaging Researchno summary yetcs cv2311.09402Google Research2 citesNov 15, 2023cs cv
- 2023Property-Aware Multi-Speaker Data Simulation: A Probabilistic Modelling Technique for Synthetic Data GenerationData2310.12371NVIDIAOct 18, 2023~106 minData
- 2023On Synthetic Data for Back Translationno summary yetcs cl2310.13675Tencent7 citesOct 20, 2023cs cl
- 2023SIMPLE SYNTHETIC DATA REDUCES SYCOPHANCY IN LARGE LANGUAGE MODELSAlignment2308.03958Aug 7, 2023~107 minAlignment
- 2022Chronological Self-Training for Real-Time Speaker Diarizationno summary yetcs sd2208.03393Google Research0 citesAug 5, 2022cs sd
- 2021Finding needles in a haystack: Sampling Structurally-diverse Training Sets from Synthetic Data for Compositional Generalizationno summary yetcs cl2109.02575AllenAI0 citesSep 6, 2021cs cl
- 2021Translate & Fill: Improving Zero-Shot Multilingual Semantic Parsing with Synthetic Datano summary yetcs cl2109.04319Google Research8 citesSep 9, 2021cs cl
- 2021Task-adaptive Pre-training and Self-training are Complementary for Natural Language Understandingno summary yetcs cl2109.06466Salesforce1 citesSep 14, 2021cs cl
- 2021Self-training with Few-shot Rationalization: Teacher Explanations Aid Student in Few-shot NLUno summary yetcs cl2109.08259Microsoft Research0 citesSep 17, 2021cs cl
- 2021Fake It Till You Make It: Face analysis in the wild using synthetic data aloneno summary yetcs cv2109.15102Microsoft Research5 citesSep 30, 2021cs cv
- 2021A Neural Acoustic Echo Canceller Optimized Using An Automatic Speech Recognizer And Large Scale Synthetic Datano summary yeteess as2106.00856Google Research0 citesJun 1, 2021eess as
- 2021Self-Training Sampling with Monolingual Data Uncertainty for Neural Machine Translationno summary yetcs cl2106.00941Tencent3 citesJun 2, 2021cs cl
- 2021SynthASR: Unlocking Synthetic Data for Speech Recognitionno summary yetcs lg2106.07803Amazon2 citesJun 14, 2021cs lg
- 2021Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensemblesno summary yetcs lg2106.15728Google Research6 citesJun 29, 2021cs lg
- 2021Synthetic Data Generation for Grammatical Error Correction with Tagged Corruption Modelsno summary yetcs cl2105.13318Google Research41 citesMay 27, 2021cs cl
- 2021Self-Training with Weak Supervisionno summary yetcs cl2104.05514Microsoft Research0 citesApr 12, 2021cs cl
- 2021Discriminative Self-training for Punctuation Predictionno summary yetcs cl2104.10339Alibaba0 citesApr 21, 2021cs cl
- 2021Training a First-Order Theorem Prover from Synthetic Datano summary yetcs ai2103.03798Google Research4 citesMar 5, 2021cs ai
- 2021Semi-Supervised Singing Voice Separation with Noisy Self-Trainingno summary yeteess as2102.07961Amazon1 citesFeb 16, 2021eess as
- 2021CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learningno summary yetcs cv2102.09559Google Research33 citesFeb 18, 2021cs cv
- 2021Asymmetric self-play for automatic goal discovery in robotic manipulationno summary yetcs lg2101.04882OpenAI21 citesJan 13, 2021cs lg
- 2020A Sharp Analysis of Model-based Reinforcement Learning with Self-Playno summary yetcs lg2010.01604Salesforce6 citesOct 4, 2020cs lg
- 2020End-to-End Synthetic Data Generation for Domain Adaptation of Question Answering Systemsno summary yetcs cl2010.06028Google Research21 citesOct 12, 2020cs cl
- 2020Learning from Mistakes: Combining Ontologies via Self-Training for Dialogue Generationno summary yetcs cl2010.00150Amazon5 citesSep 30, 2020cs cl
- 2020AutoSimulate: (Quickly) Learning Synthetic Data Generationno summary yetcs cv2008.08424Microsoft Research2 citesAug 16, 2020cs cv
- 2020Improving Object Detection with Selective Self-supervised Self-trainingno summary yetcs cv2007.09162Google Research1 citesJul 17, 2020cs cv
- 2020Unsupervised Controllable Generation with Self-Trainingno summary yetcs lg2007.09250NVIDIA0 citesJul 17, 2020cs lg
- 2020Rethinking Pre-training and Self-trainingno summary yetcs cv2006.06882Google Research49 citesJun 11, 2020cs cv
- 2020Near-Optimal Reinforcement Learning with Self-Playno summary yetcs lg2006.12007Salesforce14 citesJun 22, 2020cs lg
- 2020Capturing document context inside sentence-level neural machine translation models with self-trainingno summary yetcs cl2003.05259Google Research6 citesMar 11, 2020cs cl
- 2020Provable Self-Play Algorithms for Competitive Reinforcement Learningno summary yetcs lg2002.04017Salesforce30 citesFeb 10, 2020cs lg
- 2020Training Question Answering Models From Synthetic Datano summary yetcs cl2002.09599NVIDIA15 citesFeb 22, 2020cs cl
- 2020Semi-supervised ASR by End-to-end Self-trainingno summary yeteess as2001.09128Salesforce7 citesJan 24, 2020eess as
- 2020DP-CGAN: Differentially Private Synthetic Data and Label Generationno summary yetcs lg2001.09700Google Research1 citesJan 27, 2020cs lg
- 2020Towards Learning Multi-agent Negotiations via Self-Playno summary yetcs ro2001.10208Apple3 citesJan 28, 2020cs ro
- 2019Synthetic Datasets for Neural Program Synthesisno summary yetcs lg1912.12345Google Research7 citesDec 27, 2019cs lg
- 2019Self-training with Noisy Student improves ImageNet classificationno summary yetcs lg1911.04252Google Research241 citesNov 11, 2019cs lg
- 2019Beyond Photo Realism for Domain Adaptation from Synthetic Datano summary yetcs cv1909.01960Google Research1 citesSep 4, 2019cs cv
- 2019Learning to Generate Synthetic Data via Compositingno summary yetcs cv1904.05475Amazon6 citesApr 10, 2019cs cv
- 2018Effective Use of Synthetic Data for Urban Scene Semantic Segmentationno summary yetcs cv1807.06132NVIDIA19 citesJul 16, 2018cs cv
- 2018Learning Existing Social Conventions via Observationally Augmented Self-Playno summary yetcs ai1806.10071Meta / FAIR5 citesJun 26, 2018cs ai
- 2017A general reinforcement learning algorithm that masters chess, shogi, and Go through self-playRL Training1712.01815Dec 5, 2017~122 minRL Training